- Top of page
- Materials and methods
Objective To investigate the association between environmental exposure to livestock and incidence of diarrhoea among Vietnamese children.
Methods A population-based cohort of 353 525 individuals, living in 75 828 households in Khanh Hoa Province, Vietnam, with baseline data covering geo-referenced information on demography, socio-economic status and household animals was investigated. Geographic information system was applied to calculate the density of livestock. The data were linked to hospitalized diarrhoea cases of children under 5 years recorded at two hospitals treating patients from the area as inpatients in the study area.
Results Overall, 3116 children with diarrhoea were hospitalized during the study period. The incidence of diarrhoea hospitalization was 60.8/1000 child-years. Male gender, age <2 years, higher number of household members and lack of tap water were significantly associated with an increased risk of diarrhoea. There was no evidence that ownership of livestock increased the risk of diarrhoea. In spatial analysis, we found no evidence that a high density of any animals was associated with an increased risk of diarrhoea.
Conclusions Exposure to animals near or in households does not seem to constitute a major risk for diarrhoea in children under the age of 5 in Vietnam. Public health interventions to reduce childhood diarrhoea burden should focus on well-recognized causes such as sanitation, personal hygiene, access to adequate clean water supply and vaccination.
Objectif: Etudier l’association entre l’exposition environnementale au bétail et l’incidence de la diarrhée chez les enfants vietnamiens.
Méthodes: Une cohorte basée sur la population portant sur 353.525 personnes, vivant dans 75.828 ménages dans la province de Khanh Hoa, au Vietnam, avec des données de référence comportant les informations géo-référencées sur la démographie, le statut socioéconomique et le bétail du ménage a étéétudiée. Le Système d’Information Géographique a été appliquée pour calculer la densité du bétail. Les données ont été reliées aux cas de diarrhée hospitalisée chez les enfants de moins de 5 ans, enregistrés dans deux hôpitaux traitant les patients hospitalisés de la région.
Résultats: Au total, 3116 enfants souffrant de diarrhée ont été hospitalisés au cours de la période d’étude. L’incidence de l’hospitalisation pour diarrhée était de 60,8/1000 enfants-années. Le sexe masculin, l’âge <2 ans, un nombre plus élevé des membres du ménage et le manque d’eau de robinet étaient significativement associés à un risque accru de diarrhée. Il n’y avait aucune preuve que la possession de bétail augmentait le risque de diarrhée. Dans l’analyse spatiale, nous n’avons trouvé aucune preuve qu’une forte densité de bétail était associée à un risque accru de diarrhée.
Conclusions: L’exposition aux animaux à proximité ou dans les ménages ne semble pas constituer un risque majeur pour la diarrhée chez les enfants de moins de 5 ans au Vietnam. Les interventions de santé publique pour réduire la charge des diarrhées infantiles devraient se concentrer sur les causes bien connues telles que l’assainissement, l’hygiène personnelle, l’accès à l’eau potable et la vaccination.
Objetivo: Investigar la asociación entre la exposición ambiental al ganado y la incidencia de diarrea en niños Vietnamitas.
Métodos: Se estudió una cohorte basada en la población de 353,525 individuos que vivían en 75,828 hogares de la provincia de Khanh Hoa, Vietnam, con datos de base georeferenciados sobre demografía, estatus socioeconómico y tenencia de animales en el hogar. Se aplicaron SIG para calcular la densidad del ganado. Los datos fueron relacionados con casos de niños menores de 5 años hospitalizados por diarrea en dos hospitales del área.
Resultados: En total se hospitalizaron 3116 niños con diarrea durante el periodo de estudio. La incidencia de hospitalización por diarrea era 60.8 /1000 niños-años. El ser de género masculino, tener una edad <2 años, un mayor número de miembros en el hogar y la falta de agua corriente estaban significativamente asociados con un aumento en el riesgo de diarrea. No había evidencia de que el la tenencia de ganado aumentase el riesgo de diarrea. En un análisis espacial no se halló evidencia de que una alta densidad de animales estuviese asociada con un mayor riesgo de diarrea.
Conclusiones: La exposición de animales cercanos a o dentro de los hogares no parece constituir un riesgo mayor de diarrea en niños menores de 5 años en Vietnam. Las intervenciones de salud pública para reducir la carga por diarrea infantil deberían enfocarse en causas bien reconocidas tales como el saneamiento ambiental, la higiene personal, el acceso a una fuente adecuada de agua limpia y la vacunación.
- Top of page
- Materials and methods
In Vietnam, livestock plays an important economic role for many households. Despite rapid economic development in Vietnam, many children continue to frequently suffer from diarrhoeal diseases. Previous studies have often focused on water, sanitation and hygiene issues as risk factors for diarrhoea. The role of livestock in the transmission chain of diarrhoea pathogens is less well explored.
Salmonella and Campylobacter are frequently isolated from chicken faeces, chicken raw meat and organs (Ha & Pham 2006; Luu et al. 2006; Tran et al. 2006). Keeping poultry could contribute to transmission of diarrhoea pathogens by contaminating the environment via faeces to which children are easily exposed (Grados et al. 1988). Given high ambient temperatures, poor hygiene and insufficient food storage facilities, low-income settings in the tropics (should) provide ideal conditions for food contamination (Islam et al. 1993; Motarjemi et al. 1993; Lanata 2003).
Ruminants (e.g. cows and buffalos) and swine have been shown to contribute to the transmission of diarrhoea pathogens such as Cryptosporidium parvum and a range of Enterobacteriaceae (Germani et al. 1994). Recently, Ahmed et al. (2007) have detected rotavirus G5P strain originated from swine in a child with clinical diarrhoea in Vietnam. A study from Israel found that exposure to livestock increased the odds of giardia infection by nearly five times (Coles et al. 2009). Presence of animal faeces in the compound was found to be associated with an increased risk of diarrhoea in children in Kenya (Shivoga & Moturi 2009). However, it has also been hypothesized that animal exposure may also lead to immunity, for example, against cryptosporidium, as shown in an outbreak report (Mayne et al. 2011).
Given that a large proportion of households in Vietnam own or raise livestock, we aimed to clarify whether this activity increases the risk of diarrhoea in this setting. We explored this question through the application of geographic information system technology and analysis in the context of a well-characterized study population in south central of Vietnam and the associated risk of diarrhoea incidence in children <5 years of age.
- Top of page
- Materials and methods
During 51 241 child-years (cy) of observation under children 5 years old, there were 3116 diarrhoea episodes requiring hospital admission (60.8 per 1000 cy). Of these, 1811 diarrhoea cases (58.3%) could be linked to the census data. Figure 1 shows numbers of cases and linked cases over time. Distinct seasonal peaks of diarrhoea were detected during the mid-year dry, hot season and the 4th quarter cool, rainy season. There was no strong seasonal pattern observed among the children with diarrhoea during the study period.
Spatial analysis visually showed that diarrhoea incidence rates were lower in population dense area (Figure 2). It also identified multiple ‘hot spot’ areas with significant higher diarrhoea hospitalization rate, Z-score (≥1.96, P < 0.05) in the study area.
Figure 2. Population density (person-years/km2) (a), diarrhoea density (episodes/km2) (b), diarrhoea incidence rate (per 1000 Cy: child-years) (c), Z-score (Getis-Ord Gi*) (d) of diarrhoea in children <5 years and livestock densities per km2 (e–h) in the study area based on 250-m grid cells.
Download figure to PowerPoint
Hospitalized diarrhoea was less common in girls than in boys (Table 2). The rate was highest in children <1 year of age (77.51 per 1000 cy) and decreased by age. There was a trend towards diarrhoea being more common in households with larger number of occupants. The lack of access to tap water was significantly associated with an increased hospitalized diarrhoea. Hospitalized diarrhoea was significantly less common in urban areas than in rural areas. Crowding and parents’ education level showed no major association with diarrhoea hospital admissions, but there was a trend towards fewer admissions with increasing distance to the nearest hospital (adjusted RR 0.65, 95% CI 0.50–0.84) (Table 2).
Table 2. Univariate and multivariate analysis of risk factors for 1811 diarrhoeal episodes among 33 660 children <5 in 2005–2006
|Characteristic||Rate/1000 cy (total cy)||Crude RR (95% CI)||Adjusted RR* (95% CI)|
|Overall rate of diarrhoea||35.34 (51 241.1)||–||–|
| Male||41.88 (26 573.9)||1||1|
| Female||28.30 (24 667.2)||0.68 (0.61–0.75)‡||0.68 (0.61–0.75)‡|
| 0–11||77.51 (7728.1)||1||1|
| 12–23||73.03 (10 050.6)||0.94 (0.85–1.05)||0.95 (0.85–1.06)|
| 24–35||22.63 (11 134.3)||0.29 (0.25–0.34)‡||0.30 (0.25–0.34)‡|
| 36–47||12.02 (10 982.0)||0.16 (0.13–0.19)‡||0.16 (0.13–0.19)‡|
| 48–60||8.28 (11 346.0)||0.11 (0.09–0.13)‡||0.11 (0.09–0.13)‡|
|Number of family member|
| 0–4||29.90 (18 660.7)||1||1|
| 5–6||36.14 (15 410.9)||1.21 (1.07–1.37)†||1.11 (0.98–1.26)|
| 7+||40.54 (17 169.4)||1.36 (1.20–1.53)‡||1.16 (1.03–1.31)#|
|Composition of house|
| Concrete, Brick||35.65 (45 475.3)||1||1|
| Non-brick||32.96 (5764.5)||0.92 (0.79–1.09)||1.04 (0.88–1.22)|
| Low||30.37 (13 666.8)||0.84 (0.73–0.97)#||1.02 (0.75–1.04)|
| Middle||37.68 (25 079.4)||1.04 (0.92–1.18)||0.86 (0.73–1.02)|
| High||36.13 (12 481.2)||1||1|
| Tap||32.65 (23 890.2)||1||1|
| Others||37.70 (27 350.9)||1.15 (1.05–1.28)†||1.21 (1.06–1.38)†|
| Yes||34.60 (34 542.3)||1||1|
| No||36.85 (16 688.4)||1.07 (0.96–1.18)||1.11 (0.99–1.24)|
|Mother’s final educational level|
| Primary school||31.91 (17 175.0)||0.87 (0.76–1.00)||1.01 (0.79–1.29)|
| Secondary school||37.69 (20 988.2)||1.03 (0.91–1.18)||1.01 (0.82–1.25)|
| High school and above||36.53 (10 622.7)||1||1|
|Father’s final educational level|
| Primary school||29.98 (14 209.2)||0.84 (0.72–0.97)#||0.90 (0.68–1.20)|
| Secondary school||39.10 (18 005.7)||1.09 (0.95–1.25)||1.09 (0.86–1.38)|
| High school and above||35.83 (10 465.6)||1||1|
| Rural||36.72 (32 000.6)||1||1|
| Urban||33.06 (19 240.5)||0.90 (0.81–1.00)#||0.88 (0.76–1.01)|
|Distance to nearest hospital|
| 1–<5 km||35.61 (32 069.2)||1||1|
| 5–<10 km||37.54 (15 476.0)||1.05 (0.95–1.17)||0.99 (0.88–1.11)|
| 10–<25 km||24.18 (3639.5)||0.68 (0.53–0.86)†||0.65 (0.50–0.84)‡|
Of the households in the census area, 32% owned chicken or ducks/geese, 10% cows or buffalos, 6% swines, 50% dogs and 17% cats. Figure 2 illustrates the density of each animal in the study area. By comparing data on diarrhoea incidence rates with the distribution of domestic pets and animal livestock, we found no evidence that a high animal density area were associated with higher rate of hospital admissions. Animal ownership in general was more common in rural than urban Khanh Hoa (65%vs. 28% for dogs, 46%vs. 3% for chicken, 10%vs. 1% for swine and 16%vs. 0.1% for cows). Excluding households with no animals, the mean number of animals per household was 4.6 (range 1–80) for swine, 4.7 (range 1–31) for buffalos, 1.5 (range 1–13) for dogs, 1.2 (range 1–61) for cats, 3.4 (range 1–80) for cows, 15.7 (range 1–7000) for chicken and 21.5 (1–6000) for ducks and geese. Livestock ownership was lower in households in the high-wealth category than in the low- and middle-wealth category (which owned similar numbers of the different animal types). Comparing low- and middle- with the high-wealth category, swine ownership was 7%vs. 3%, cow ownership 13%vs. 2%, chicken ownership 34%vs. 10%, duck/geese ownership 19%vs. 4%, and dog or cat ownership 36%vs. 25% (data not shown).
In univariate and multivariate analysis, children living in households with poultry, ruminants, swine, dogs or cats had no increased risk of diarrhoea hospital admissions (Table 3). We also found no evidence for a dose-response relationship; children in households in the upper quintile of animal density neighbourhood were at no significant higher risk of diarrhoea admission compared with children in the lowest quintile (Table 3).
Table 3. Univariate and multivariate analysis for 1811 diarrhoea episodes among 33 660 eligible subjects 2005–2006
|Characteristic||Rate/1000PYs (total cy)||Crude RR (95% CI)||Adjusted RR* (95% CI)|
| Yes||39.80 (5903.9)||1.15 (0.99–1.33)||1.15 (0.99–1.34)|
| No||34.76 (45 337.2)||1||1|
| Yes||34.95 (3175.8)||0.99 (0.80–1.22)||0.90 (0.72–1.11)|
| No||35.37 (48 065.3)||1||1|
| Yes||36.22 (16 425.4)||1.04 (0.93–1.15)||0.96 (0.85–1.09)|
| No||34.93 (34 815.7)||1||1|
| Yes||37.82 (24 777.3)||1.15 (1.04–1.26)†||1.10 (0.99–1.23)|
| No||33.03 (26 463.8)||1||1|
| Yes||39.00 (7973.7)||1.13 (0.99–1.28)||1.06 (0.93–1.22)|
| No||34.67 (43 267.4)||1||1|
|Poultry (chicken & duck) density quintile (mean poultry per km2)|
| 1st quintile (59)||34.74 (9527.5)||1||1|
| 2nd quintile (306)||36.47 (9925.8)||1.05 (0.89–1.24)||1.04 (0.87–1.24)|
| 3rd quintile (1113)||29.97 (10 109.4)||0.86 (0.72–1.03)||0.82 (0.68–0.98)|
| 4th quintile (3506)||38.28 (10 318.1)||1.10 (0.94–1.29)||0.99 (0.81–1.21)|
| 5th quintile (13850)||34.74 (9527.5)||1.10 (0.93–1.29)||0.94 (0.78–1.15)|
|Dog and cat density quintile (mean dog/cats per km2)|
| 1st quintile (217)||35.27 (10 605.2)||1||1|
| 2nd quintile (514)||35.30 (10 284.6)||1.00 (0.86–1.17)||0.99 (0.85–1.15)|
| 3rd quintile (815)||37.39 (10 296.0)||1.06 (0.90–1.25)||1.03 (0.88–1.22)|
| 4th quintile (1238)||34.72 (9676.3)||0.98 (0.84–1.16)||1.00 (0.84–1.18)|
| 5th quintile (2092)||34.99 (9546.6)||0.99 (0.84–1.18)||1.02 (0.84–1.22)|
|Swine density quintile (mean swine per km2)|
| 1st quintile (3)||36.26 (10 343.0)||1||1|
| 2nd quintile (50)||33.57 (9949.8)||0.93 (0.78–1.10)||0.89 (0.75–1.05)|
| 3rd quintile (116)||34.68 (10 209.0)||0.96 (0.81–1.12)||0.90 (0.77–1.05)|
| 4th quintile (217)||38.39 (9923.9)||1.06 (0.90–1.25)||0.98 (0.83–1.15)|
| 5th quintile (646)||34.86 (9983.0)||0.96 (0.82–1.13)||0.88 (0.75–1.04)|
|Ruminant (cow and buffalo) density quintile (mean ruminants per km2)|
| 1st quintile (0)||33.81 (17 364.4)||1||1|
| 2nd quintile (1)||39.28 (1985.8)||1.05 (0.58–1.91)||1.07 (0.85–1.34)|
| 3rd quintile (30)||34.92 (9967.1)||0.79 (0.51–1.20)||0.96 (0.80–1.15)|
| 4th quintile (137)||36.51 (10 545.4)||0.92 (0.62–1.38)||0.97 (0.80–1.17)|
| 5th quintile (489)||37.36 (10 546.1)||1.33 (0.92–1.93)||1.02 (0.84–1.25)|
|Total animal density quintile (mean animals per km2)|
| 1st quintile (743)||33.37 (10 009.4)||1||1|
| 2nd quintile (1687)||34.43 (9 788.2)||1.03 (0.87–1.23)||1.03 (0.86–1.22)|
| 3rd quintile (2682)||36.59 (9920.9)||1.10 (0.92–1.30)||1.05 (0.89–1.24)|
| 4th quintile (4763)||35.29 (10 143.4)||1.06 (0.89–1.26)||0.99 (0.83–1.18)|
| 5th quintile (15474)||37.93 (10 546.9)||1.14 (0.96–1.34)||1.00 (0.84–1.20)|
Because of the strong association between tap water availability and diarrhoea (Table 2), we investigated the relationship between animal ownership and an increased risk of diarrhoea in households without tap water, where the risk of water contamination with pathogens may be greatest. We found no evidence for effect modification (interaction) based on the presence or absence of household tap water. There was no increased risk of diarrhoea associated with any type of animal in households without tap water: poultry, RR 0.96 (95% CI: 0.85–1.09); swines, RR 0.90 (95% CI: 0.72–1.11); ruminants, RR 1.15 (95% CI: 0.99–1.34); dogs, RR 1.10 (95% CI: 0.99–1.23); cats, RR 1.06 (95% CI: 0.93–1.22) (Table 3).
We further investigated an association between animal ownership and diarrhoea hospitalizations by analysing the study population separately for (i) rural areas and urban, (ii) diarrhoea incidence in 2005 and 2006, (iii) diarrhoea cases during the peaks and not during the peaks, and (iv) diarrhoea cases inside ‘hot spot’ areas. However, none of these factors was found to modify the lack of an association between animal ownership and diarrhoea.
- Top of page
- Materials and methods
We found no evidence that children under the age of 5 years, living closely with livestock and other animals, have an increased rate of hospitalization with diarrhoea. Extensive subgroup analyses (e.g. water access, rural/urban, year, high-incidence/low-incidence periods) revealed no additional insights.
At population level, human-to-human transmission is likely to be the predominant mode of transmission. Animals may constitute an important source of diarrhoea pathogens, even if (as our data suggest) exposure to animals seems to pose few immediate risks in the directly exposed population of children. Zoonotic transmission from animals to humans (for example, of cryptosporidia) may occur, perhaps even frequently, causing subsequent human-to-human transmission. We cannot exclude that a separation of animals from humans may reduce zoonotic infections and epidemics in humans at a later stage.
Our findings do not eliminate the possibility that contamination of animal food products with enteric pathogens may contribute substantially to the burden of diarrhoea. The importance of adequate water supply was confirmed by our analysis, although we are unable to determine whether the beneficial effect of tap water supply was attributed to improved water quantity, quality or both. As rotavirus is one of the major causes of childhood diarrhoea, introduction of rotavirus vaccine can be highly effective in reducing hospitalizations associated with rotavirus diarrhoea in children (Zaman et al. 2010; Anh et al. 2011).
Thus, control efforts could include addressing issues of food hygiene, sanitation, personal hygiene, access to adequate clean water supply and vaccination.
Our analysis may be limited by the potential of residual confounding in particular from socio-economic characteristics, access to health care and analysis limited to hospitalized cases. Children admitted to hospital are more likely to have diarrhoea of greater severity. This does not need to be a weakness of this analysis because severe cases are those of highest public health interest. We found that education and wealth level were not strongly associated with diarrhoea. Higher use of health services in educated, wealthy individuals (usually at low risk of diarrhoea) may have masked an actually higher risk of diarrhoea in less-educated and less-wealthy families. Free health care for children under the age of 6 years has been introduced in the study area since October 2005, that is, for most of the study period, healthcare costs should not have been a barrier to healthcare seeking. We found that this fee exemption had little impact on actual admission rates.
We were not able to link 40% of cases to the census, possibly due to diarrhoea cases among the mobile sections of the population and newly born babies after the census, who were not part of the cohort. It seems unlikely that animal ownership is related to chance of a hospitalization being linked to the census data, which would introduce bias.
We found a strong relationship between hospitalized diarrhoea and distance to the nearest hospital. This might be a result of healthcare seeking behaviour, which was consistent with findings from another study conducted in Khanh Hoa in 2002 (Ali et al. 2005b). Analytically, it is not always easy to adjust for distance to the hospital. We used a simple straight line distance, ignoring actual travel distances. In a sensitivity analysis, we restricted the calculations to people living within 10 km of the hospitals, but found no material change in the results.
Another limitation could be that the animal data were collected in a cross-sectional survey but were later applied for the whole study period while animal numbers may be changing over time. In a sensitivity analysis, we restricted the calculations to the year 2006 when the census was carried out, and for which the animal data should be most accurate. The results were very similar to the analysis of all years. We used a kernel density method to calculate animal density. In a sensitivity analysis, we tried an alternative measure by counting the number of animals within a radius of 100 around each household. Using these measures instead did not greatly alter the results shown in Table 3.
Based on these data, we conclude that exposure to animals near or in households do not seem to constitute a major risk for diarrhoea in children under the age of 5. Our analysis provides indirect support for focusing control efforts on human-to-human transmission, for example, by improved water access, sanitation, hygiene (Cairncross et al. 2010) or vaccination against specific pathogens such as rotavirus.
- Top of page
- Materials and methods
We are grateful to the study participants, Khanh Hoa Health Service, especially Dr Truong Tan Minh and directorate boards of Khanh Hoa General and Ninh Hoa District Hospitals. We are grateful to the participating commune health centres for their assistance during the fieldwork of this study. We thank Drs Kensuke Goto, Susumu Tanimura and Toru Matsubayashi of Nagasaki University, Dr Paul Kilgore of the International Vaccine Institute, Seoul, Korea, for their help on develop the study protocol and data analysis. This work is supported by the Japan Initiative for Global Research Network on Infectious Diseases (J-GRID), Ministry of Education, Culture, Sports, Science and Technology, Japan. We thank Dr Masahiro Hashizume for critical comments on the manuscript.